WEZZEL NEWS

Knowing but Not Saying: Preventing Factual Access Failures in LLM SFT via Recall-Anchored Distillation

Researchers have identified a problem with large language models (LLMs) called factual access failur...

Automated Trajectory Evaluation for Mobile Agents via Step-Level Consequence Reasoning and Aggregation

Researchers have developed a new method for evaluating the performance of mobile agents, such as rob...

Dynamic Context Scheduling: Learning Beyond the Static Universe

Researchers have developed a new approach called dynamic context scheduling, which involves training...

SPARC: Single-Pass Scaling for Motion Forecasting with Conformal Bayesian Last Layers

Researchers have developed SPARC, a new method for motion forecasting that estimates uncertainty in ...

Neuro-Geospatial Modelling of EEG Affective States Using Literature-Informed Environmental Context

Researchers have developed a method to combine brain activity data from EEG with...

Certified Multi-Turn Robustness for LLM Safety via Compositional Bounds and Safety Persistence

Researchers have proposed a method to improve the safety of large language model...

Prediction certification cannot replace explanation certification: a competence envelope for trustworthy AI under compound stress

Researchers have found that relying solely on prediction-based certification for...

Foundation Models for Partial Causal Identification

Researchers have developed a new approach to identifying the cause-and-effect re...